The TEI workgroup originally charged with producing the means for encoding the 
shelfmark (and all its appurtenances) planned on:

<!ELEMENT msIdentifier
 	(country | region | settlement | institution | repository
 	| collection | idno | altName)+ >

The reasoning was that much as one might like to have all this information, 
there are times when it simply is not available or is available in all 
too many ways:

--a manuscript sold at auction and now lost to view
--a manuscript whose owner claims two legal residences (cf. the Schoeyen 
Collection, legally designated as London/Oslo)
--a manuscript legally owned by more than one institution (the Newberry 
Library in Chicago that makes joint purchases with other libraries, e.g. that 
of Notre Dame)
--a manuscript known to the encoder only by its nickname, "The Grimani 
Breviary" or known to scholarship by a nickname that refers to only its most 
important (??) aspect, e.g. "The Berlin Grosseteste" (as if there were only 
one manuscript of Grosseteste in Berlin, and as if there were only one 

The thinking above planned not just for produced-to-fit modern catalogues 
of manuscripts, but also for legacy data in the shape of older catalogues; 
it also envisioned the encoding of references to a manuscript's stemma 
(e.g. the "P" manuscript of Livy), a scholar's direct encoding of his own 
material even for private use, and the encoding of journal articles, in 
particular those published in the 19th century, where citations of 
manuscripts are very iffy (of the sort that reads, "a manuscript of Alcuin 
shown to me by a friend").

As Sebastian Rahtz points out, by now we can only deal with the results of 
the work group who had been assigned to reconcile what the TEI work group 
did with what the MASTER group did.  Or we can customize.

\Consuelo Dutschke

Date: Fri, 14 Sep 2007 23:49:03 +0100
From: Sebastian Rahtz <[log in to unmask]>
To: [log in to unmask]
Subject: Re: trouble with msIdentifier

If I understand this thread and the MS module aright,
then I believe that the mandatory <repository> business is intentional,
properly implemented, and the outcome of quite
detailed and specific discussion by that work group.

Given that we are on course to release TEI P5 1.0 in about 6
weeks time, there really is not an opportunity to
revisit such decisions at this stage. We are
scheduled to freeze technical code any moment now...

Of course, there is life after P5 1.0; and there is